Model - Logistics¶

In [2]:

            
                Copied!
                
from pandas_profiling import ProfileReport
import pandas as pd
import numpy as np
from pandas_profiling import ProfileReport
import pandas as pd
import numpy as np

c:\Users\ericl\miniconda3\lib\site-packages\tqdm\auto.py:22: TqdmWarning: IProgress not found. Please update jupyter and ipywidgets. See https://ipywidgets.readthedocs.io/en/stable/user_install.html
  from .autonotebook import tqdm as notebook_tqdm

EDA¶

In [55]:

            
                Copied!
                
df = pd.read_csv('train.csv')
df = pd.read_csv('train.csv')

In [11]:

            
                Copied!
                
df.head()
df.head()

Out[11]:

	raw_row_number	location	county_name	subject_age	subject_race	subject_sex	officer_id_hash	department_name	type	arrest_made	...	outcome	frisk_performed	search_conducted	search_person	search_vehicle	reason_for_stop	raw_Ethnicity	raw_Race	raw_action_description	date_time
0	12511107	NaN	forsyth county	18.0	white	male	f2f6b08c97	Winston-Salem Police Department	vehicular	False	...	citation	False	False	False	False	Speed Limit Violation	N	W	Citation Issued	2010-12-05 01:51:24
1	5439683	raleigh	wake county	25.0	hispanic	male	1e3fa73f20	Raleigh Police Department	vehicular	False	...	warning	False	True	True	True	Vehicle Regulatory Violation	H	W	Verbal Warning	2005-09-25 03:40:00
2	18674698	charlotte area	mecklenburg county	30.0	black	female	59a754eb04	Charlotte-Mecklenburg Police Department	vehicular	False	...	warning	False	False	False	False	Speed Limit Violation	N	B	Verbal Warning	2014-11-15 02:00:00
3	12600300	charlotte area	mecklenburg county	21.0	white	male	0dc507ea69	Charlotte-Mecklenburg Police Department	vehicular	False	...	warning	False	False	False	False	Vehicle Regulatory Violation	N	W	Verbal Warning	2011-01-23 00:16:00
4	6035053	NaN	durham county	38.0	black	female	91822b2dfe	Durham Police Department	vehicular	False	...	citation	False	False	False	False	Speed Limit Violation	N	B	Citation Issued	2006-06-18 10:17:17

5 rows × 22 columns

Feature Engineering¶

In [7]:

            
                Copied!
                
profile = ProfileReport(df, title="Police Stop Profiling Report")
profile
profile = ProfileReport(df, title="Police Stop Profiling Report")
profile

Summarize dataset: 100%|██████████| 41/41 [06:19<00:00,  9.25s/it, Completed]                               
Generate report structure: 100%|██████████| 1/1 [00:07<00:00,  7.29s/it]
Render HTML: 100%|██████████| 1/1 [00:01<00:00,  1.73s/it]

Out[7]:

In [ ]:

In [56]:

            
                Copied!
                
#Getting rid of unneccessary columns 
df["outcome"].replace(['warning', 'citation', 'arrest'], [0, 1, 2], inplace=True)
df.drop(columns=['date_time', 'department_name', 'county_name', 'type', 'raw_Race', 'raw_Ethnicity', 'raw_action_description', 'subject_race', 'location', 'raw_row_number', 'officer_id_hash', 'arrest_made', 'citation_issued', 'warning_issued'], inplace=True)
#Getting rid of unneccessary columns 
df["outcome"].replace(['warning', 'citation', 'arrest'], [0, 1, 2], inplace=True)
df.drop(columns=['date_time', 'department_name', 'county_name', 'type', 'raw_Race', 'raw_Ethnicity', 'raw_action_description', 'subject_race', 'location', 'raw_row_number', 'officer_id_hash', 'arrest_made', 'citation_issued', 'warning_issued'], inplace=True)

In [57]:

            
                Copied!
                
                    
                    
                
                

        
#Change reason for stop and boolean columns to numerical
df = pd.concat([df, df['reason_for_stop'].str.get_dummies()], axis=1)
df.drop(columns=['reason_for_stop'], inplace=True)
df['subject_sex'].replace({'male': 1, 'female': 0}, inplace=True)
df['frisk_performed'].replace({True: 1, False: 0}, inplace=True)
df['search_conducted'].replace({True: 1, False: 0}, inplace=True)
df['search_person'].replace({True: 1, False: 0}, inplace=True)
df['search_vehicle'].replace({True: 1, False: 0}, inplace=True)
#Change reason for stop and boolean columns to numerical
df = pd.concat([df, df['reason_for_stop'].str.get_dummies()], axis=1)
df.drop(columns=['reason_for_stop'], inplace=True)
df['subject_sex'].replace({'male': 1, 'female': 0}, inplace=True)
df['frisk_performed'].replace({True: 1, False: 0}, inplace=True)
df['search_conducted'].replace({True: 1, False: 0}, inplace=True)
df['search_person'].replace({True: 1, False: 0}, inplace=True)
df['search_vehicle'].replace({True: 1, False: 0}, inplace=True)

In [76]:

            
                Copied!
                
ProfileReport(df, title='Useful Profile Report')
ProfileReport(df, title='Useful Profile Report')

Summarize dataset: 100%|██████████| 32/32 [09:45<00:00, 18.29s/it, Completed]                                     
Generate report structure: 100%|██████████| 1/1 [00:09<00:00,  9.28s/it]
Render HTML: 100%|██████████| 1/1 [00:01<00:00,  1.35s/it]

Out[76]:

In [66]:

            
                Copied!
                
                    
                    
                
                

        
#read test files and also perform the same changes
test = pd.read_csv('test.csv')
test["outcome"].replace(['warning', 'citation', 'arrest'], [0, 1, 2], inplace=True)
test.drop(columns=['date_time', 'department_name', 'county_name', 'type', 'raw_Race', 'raw_Ethnicity', 'raw_action_description', 'subject_race', 'location', 'raw_row_number', 'officer_id_hash', 'arrest_made', 'citation_issued', 'warning_issued'], inplace=True)
test = pd.concat([test, test['reason_for_stop'].str.get_dummies()], axis=1)
test.drop(columns=['reason_for_stop'], inplace=True)
test['subject_sex'].replace({'male': 1, 'female': 0}, inplace=True)
test['frisk_performed'].replace({True: 1, False: 0}, inplace=True)
test['search_conducted'].replace({True: 1, False: 0}, inplace=True)
test['search_person'].replace({True: 1, False: 0}, inplace=True)
test['search_vehicle'].replace({True: 1, False: 0}, inplace=True)
#read test files and also perform the same changes
test = pd.read_csv('test.csv')
test["outcome"].replace(['warning', 'citation', 'arrest'], [0, 1, 2], inplace=True)
test.drop(columns=['date_time', 'department_name', 'county_name', 'type', 'raw_Race', 'raw_Ethnicity', 'raw_action_description', 'subject_race', 'location', 'raw_row_number', 'officer_id_hash', 'arrest_made', 'citation_issued', 'warning_issued'], inplace=True)
test = pd.concat([test, test['reason_for_stop'].str.get_dummies()], axis=1)
test.drop(columns=['reason_for_stop'], inplace=True)
test['subject_sex'].replace({'male': 1, 'female': 0}, inplace=True)
test['frisk_performed'].replace({True: 1, False: 0}, inplace=True)
test['search_conducted'].replace({True: 1, False: 0}, inplace=True)
test['search_person'].replace({True: 1, False: 0}, inplace=True)
test['search_vehicle'].replace({True: 1, False: 0}, inplace=True)

In [71]:

            
                Copied!
                
#Seperate data for modeling
X_train = df.drop('outcome', axis=1)
X_test = test.drop('outcome', axis=1)
Y_train = df['outcome']
Y_test = test['outcome']
#Seperate data for modeling
X_train = df.drop('outcome', axis=1)
X_test = test.drop('outcome', axis=1)
Y_train = df['outcome']
Y_test = test['outcome']

In [73]:

            
                Copied!
                
from sklearn.linear_model import LogisticRegression

#Make the model
model = LogisticRegression(solver = 'liblinear')
model.fit(X_train,Y_train)
Y_pred = model.predict(X_test)
from sklearn.linear_model import LogisticRegression

#Make the model
model = LogisticRegression(solver = 'liblinear')
model.fit(X_train,Y_train)
Y_pred = model.predict(X_test)

Result¶

In [75]:

            
                Copied!
                
#Accuracy
np.mean(Y_test==Y_pred)
#Accuracy
np.mean(Y_test==Y_pred)

Out[75]:

0.661227829379674

Not Very Accurate!

In [78]:

            
                Copied!
                
#Try with more relavent columns
X_train = df[['search_conducted', 'Vehicle Equipment Violation', 'Driving While Impaired', 'Safe Movement Violation', 'Speed Limit Violation']]
X_test = test[['search_conducted', 'Vehicle Equipment Violation', 'Driving While Impaired', 'Safe Movement Violation', 'Speed Limit Violation']]
Y_train = df['outcome']
Y_test = test['outcome']
#Try with more relavent columns
X_train = df[['search_conducted', 'Vehicle Equipment Violation', 'Driving While Impaired', 'Safe Movement Violation', 'Speed Limit Violation']]
X_test = test[['search_conducted', 'Vehicle Equipment Violation', 'Driving While Impaired', 'Safe Movement Violation', 'Speed Limit Violation']]
Y_train = df['outcome']
Y_test = test['outcome']

In [79]:

            
                Copied!
                
#Make the model
model = LogisticRegression(solver = 'liblinear')
model.fit(X_train,Y_train)
Y_pred = model.predict(X_test)

#Accuracy
np.mean(Y_test==Y_pred)
#Make the model
model = LogisticRegression(solver = 'liblinear')
model.fit(X_train,Y_train)
Y_pred = model.predict(X_test)

#Accuracy
np.mean(Y_test==Y_pred)

Out[79]:

0.6585383071749138